ISO/PWI TS 21947

Machine Learning Application to Surface Analysis Data. Part I: Mass Spectrum and Imaging

Apstrakt

This technical report (TS) provides guidelines of machine learning application to mass spectrometry data including spectra and mass images. Machine learning includes numerical analysis methods using computers such as multivariate analysis, artificial neural network-based methods and combination of multiple learning methods, and is generally classified into three categories, unsupervised learning, supervised learning, and reinforcement learning. This TS gives explanations of basic terms regarding machine learning and shows suitable methods depending on the analysis purposes and necessary procedures to data preparation for machine learning. This TS does not strictly regulate learning methods for mass spectrometry data because new learning methods and modified methods are being developed.